Smart Algorithms

What would analytics be without clever algorithms that work in the background and tell you as a user in advance which search terms it would be better to take a closer look at. We use two different machine learning algorithms for this purpose, each of which provides you with different tips. The trend and outlier analysis is broken down into two parts. The trend analysis is represented by a small grey arrow next to each KPI. The arrow points either upwards for an upward trend or downwards for a downward trend. The respective intensity of the algorithm is visualised by the intensity of the grey tone. In the following example we have looked at the top search terms of our channel and notice that there should be an upward trend for the search term "fan":

If we pull up the search term and select the comparison period in the top right corner, we can easily see the trend. The dotted line represents the previous period:

The search volume of the current week is significantly higher!

The outlier analysis works in the same way, always calculating whether a search term has an unusual outlier compared to the lead time. Here we have sorted the search terms in descending order of outliers:

In a direct comparison of the preliminary period in the graph view, the outliers can be seen at a glance. The dotted line shows the previous week:

The outlier analysis is particularly helpful in identifying bots and robots that submit an extremely large number of search queries at certain times or click through many products but do not buy anything. Here you should look at the sessions again in detail and compare the transmitted session ID with the one in the server logs in order to blacklist the IP if necessary.